AN ABSTRACT OF THE THESIS OF Kelly Maren Kibler for the degree of Master of Science in Forest Engineering presented on June 28, 2007. Title: The Influence of Contemporary Forest Harvesting on Summer Stream Temperatures in Headwater Streams of Hinkle Creek, Oregon. Abstract approved:
Arne E. Skaugset
Stream temperature is a water quality parameter that directly influences
the quality of aquatic habitat, particularly for cold-water species such as Pacific
salmonids. Forest harvesting adjacent to a stream can increase the amount of
solar radiation the stream receives, which can elevate stream temperatures
and impair aquatic habitat. Oregon Forest Practice Rules mandate that forest
operators leave Riparian Management Areas (RMAs) adjacent to streams in
order to minimize the water quality impacts from forest harvesting. However,
RMAs that contain overstory merchantable conifers are not required for small
non-fish-bearing streams in Oregon, thus there is potential for increases in
stream temperature to occur in headwater streams after harvesting. There is
concern that increases in stream temperatures and changes to onsite
processes in non-fish-bearing, headwater streams may propagate
downstream and impair habitat in fish-bearing streams. The objectives of the
following work are to assess the effects of contemporary forest management
practices on stream temperatures of small non-fish-bearing headwater
streams and to develop new knowledge regarding the physical processes that
control reach-level stream temperature patterns.
Summer stream temperatures were measured for five years in six
headwater streams in the Hinkle Creek basin in southern Oregon. After four
years, four of the streams were harvested and vegetated RMAs were not left
between the streams and harvest units. The watersheds of the two remaining
streams were not disturbed. Post-harvest stream temperatures were
monitored for one year in all six streams. Each harvested stream was paired
with one unharvested stream and regression relationships for maximum,
minimum and mean daily stream temperatures were developed. Changes to
temperatures of harvested streams were detected by comparing the mean
pre-harvest regression relationship to the mean post-harvest relationship.
Change detection analyses that considered the mean response among all four
harvested streams indicated that maximum daily stream temperatures did not
increase after harvesting, but that minimum and mean daily temperatures
decreased significantly after harvesting. Additionally, diel stream temperature
fluctuations were significantly greater one year after harvesting.
Pre- and post-harvest surveys of canopy closure in the harvested and
unharvested streams were completed in order to compare levels of stream
shading before and after harvest. The post-harvest survey quantified canopy
closure from remaining overstory vegetation as well as from logging slash that
partially covered the harvested streams. The surveys indicated that mean
overstory canopy closure in the harvested streams decreased by 84% as a
result of the harvest, but as the logging slash provided considerable cover,
total canopy closure decreased by only 20%. It is possible that the logging
slash effectively attenuated solar radiation and prevented extreme
temperature increases in the harvested streams. However, it is likely that
streamflow increased after harvesting and that the increased streamflow also
prevented increases to maximum temperatures and contributed to lower
minimum and mean stream temperatures.
Copyright by Kelly Maren Kibler June 28, 2007
All Rights Reserved
The Influence of Contemporary Forest Harvesting on Summer Stream Temperatures in Headwater Streams of Hinkle Creek, Oregon
by Kelly Maren Kibler
A THESIS
submitted to
Oregon State University
in partial fulfillment of the requirements for the
degree of
Master of Science
Presented June 28, 2007 Commencement June 2008
Master of Science thesis of Kelly Maren Kibler presented on June 28, 2007.
APPROVED:
Major Professor, representing Forest Engineering
Head of the Department of Forest Engineering
Dean of the Graduate School
I understand that my thesis will become part of the permanent collection of
Oregon State University libraries. My signature below authorizes release of
my thesis to any reader upon request.
Kelly Maren Kibler, Author
ACKNOWLEDGEMENTS
There are no words that adequately express my gratitude to all of the
people who have contributed to the following work. To begin with, I would like
to thank all of the people who have spent their summers living less than
luxuriously in southern Oregon helping to collect data for this project. Three
years of data were collected for this project before I even joined the Hinkle
team and I am severely in debt to those who came before me and to those
who put up with me when I finally did show up. Thank you Amy Simmons,
Nicolas Zegre, Matt Meadows, Tim Otis, Jennifer Fleuret, Dennis Feeney, Tim
Royer, Kelly Hoefer, Alison Collette and Kent Smith. Once the data had been
collected, I had no idea what to do with it and so next I must thank Manuela
Huso and Lisa Ganio for leading me down the path towards defensible
statistics and enduring my statistical ignorance with minimal cringing. Thank
you to my committee members for your dedication and direction. Most of all, I
would like to thank my advisor, Arne Skaugset for his inspiration and
leadership. Every time you read something in the following pages that makes
you think, Hey, thats neat!, thank Arne because it was probably his idea.
The entire Hinkle Creek study would not have been possible if not for the
vision and cooperation of the people at Roseburg Forest Products and I
applaud them for taking a risk for the sake of advancing our knowledge of
watershed management.
Finally, I would like to thank my family, Dad, Mom, Dane, Jess, Carole
and Gary, who always found ways to love, support and encourage me, even
from 3,000 miles away. Special thanks to Mom and Dad for inspiring me to
always challenge myself. At this time, I have to extend the greatest gratitude
to my best friend and the love of my life, Benjamin Washabaugh Kibler. Thank
you Ben for taking the gamble and rollin out West with me and for your
endless support and unconditional love. As a gift to you, I would like to
dedicate my work to Grandma Betty, a true renaissance woman.
TABLE OF CONTENTS
Page Chapter I: Introduction..................................................................................... 1
Justification................................................................................................... 1 Literature review ........................................................................................... 2
Physical controls to stream temperature................................................... 2 Physical effects of stream temperature..................................................... 8 Ecological effects of stream temperature.................................................. 9 Stream temperature and forestland management................................... 14
Chapter II: The influence of contemporary forest harvesting on summer stream temperatures in headwater streams of Hinkle Creek, Oregon............ 19
Introduction................................................................................................. 19 Methods...................................................................................................... 21
Site description ....................................................................................... 21 Study design ........................................................................................... 23 Harvesting treatment............................................................................... 23 Stream temperature data collection ........................................................ 25 Canopy closure data collection ............................................................... 25 Data analysis .......................................................................................... 28
Maximum, minimum and mean daily stream temperatures ................. 28 Diel temperature fluctuation................................................................. 31 Greatest annual seven-day moving mean of the maximum daily temperature ......................................................................................... 33 Cumulative degree days...................................................................... 34 Canopy closure.................................................................................... 34
Results........................................................................................................ 34
Maximum, minimum and mean daily stream temperatures..................... 34 Diel temperature fluctuation .................................................................... 46 Greatest annual seven-day moving mean of the maximum daily temperature............................................................................................. 48 Cumulative degree days ......................................................................... 49 Canopy closure ....................................................................................... 49
Discussion .................................................................................................. 54
Analysis .................................................................................................. 54 Maximum, minimum and mean daily stream temperatures..................... 60 Diel temperature fluctuation .................................................................... 62 Degree days............................................................................................ 63
TABLE OF CONTENTS (Continued)
Page
Experimental design and individual stream reach analysis ..................... 64 Canopy closure ....................................................................................... 66 Further explanation of results.................................................................. 72 Future considerations for stream temperatures in Hinkle Creek ............. 75 Hindsight ................................................................................................. 78
Chapter III: Conclusions................................................................................ 80
Conclusions ................................................................................................ 80 References ................................................................................................. 83
Appendix A..................................................................................................... 91
LIST OF FIGURES
Figure Page
1.1 Daily patterns of net radiation (Nr), evaporation (E) and convection (H) for a shaded (a) and unshaded (b) stream (Brown 1969)............................4
2.1 Hinkle Creek study area. Black points represent approximate locations of temperature data loggers, flumes, transition points between fish-bearing and non-fish-bearing stream designations and downstream limits of timber harvesting in harvested streams. ........................................... 22
2.2. The locations of flumes and reaches surveyed for canopy closure in 2004 and 2006. The number of sampling points taken during the 2006 survey is displayed by each reach. The number of sampling points taken during the 2004 survey was equal or greater than the 2006 survey sample size for each reach. ........................................................................... 27
2.3a. Regressions of maximum daily stream temperatures in harvested streams versus unharvested streams. Each stream pair is shown individually. 95% prediction limits are around pre-harvest data..................... 40
2.3b. Regressions of minimum daily stream temperatures in harvested streams versus unharvested streams. Each stream pair is shown individually. 95% prediction limits are around pre-harvest data..................... 41
2.3c. Regressions of mean daily stream temperatures in harvested streams versus unharvested streams. Each stream pair is shown individually. 95% prediction limits are around pre-harvest data..................... 42
2.4a. Regressions of maximum daily stream temperatures in harvested streams versus unharvested streams for each stream pair and year illustrate variability of the harvested-unharvested relationship before and after harvest. Mean pre- and post-harvest regressions illustrate comparisons made by the change detection model. Vertical dashed line indicates mean intercept. ............................................................................... 43
LIST OF FIGURES (Continued)
Figure Page
2.4b. Regressions of minimum daily stream temperatures in harvested streams versus unharvested streams for each stream pair and year illustrate variability of the harvested-unharvested relationship before and after harvest. Mean pre- and post-harvest regressions illustrate comparisons made by the change detection model. Vertical dashed line indicates mean intercept. ............................................................................... 44
2.4c. Regressions of daily mean stream temperatures in harvested streams versus unharvested streams for each stream pair and year illustrate variability of the harvested-unharvested relationship before and after harvest. Mean pre- and post-harvest regressions illustrate comparisons made by the change detection model. Vertical dashed line indicates mean intercept. ............................................................................... 45
2.5. Diel fluctuation in stream temperature for every stream pre- and post-harvest. DeMersseman and Myers are unharvested............................. 47
2.6. Annual maximum seven-day mean stream temperature in all streams, pre- and post-harvest. Error bars display one standard deviation from the mean of four pre-harvest years. *Myers and DeMersseman are unharvested. .................................................................... 49
2.7 Cumulative degree days in four harvested and one unharvested stream for 2004, 2005 and 2006. Degree-day accumulation begins each year on March 1 and ends on September 30. ................................................ 51
2.8. Error analysis: Percent canopy closure for all unharvested reaches. Error bars are one standard deviation of the mean. Final group represents mean values across all unharvested reaches............................... 52
2.9. Percent canopy closure for uncut and clearcut portions of the Clay DS reach which was harvested in 2001. Error bars are one standard deviation of the mean. .................................................................................... 52
2.10. Percent canopy closure in harvested reaches. Error bars are one standard deviation of the mean. Final group represents mean values across all harvested reaches.......................................................................... 53
LIST OF FIGURES (Continued)
Figure Page
2.11. Comparison of lines with same slopes but different intercepts. ............ 56
2.12. Comparison of regression lines with different slopes but same intercept. Slopes are greater than one, equal to one and less than one. ...... 57
2.13. Comparison of lines with different slopes and different intercepts. Slopes are greater than one, equal to one and less than one; intercepts are -1, 0 and 1................................................................................................ 58
LIST OF APPENDIX FIGURES
Figure Page
A1-A6. The percent canopy closure before harvest (2004) and after harvest (2006) measured using a spherical densitometer and a digital camera (2006). The x-axis is the location of the sampling points along the streams longitudinal profile. The zero position marks the downstream boundary of the harvest unit. The mean and standard deviations of percent canopy closure after harvest in harvested reaches are shown for data collected using a spherical densitometer and a digital camera. .......................................................................................................... 94
A1- Fenton Creek........................................................................................... 94
A2- Clay Creek............................................................................................... 95
A3- Russell Creek .......................................................................................... 95
A4- BB Creek ................................................................................................. 96
A5- Myers Creek ............................................................................................ 96
A6- DeMersseman Creek............................................................................... 97
A7. Daily minimum and maximum stream temperatures plotted in time series for Fenton Creek 2002-2006 and Myers Creek (unharvested) 2005. .............................................................................................................. 98
LIST OF TABLES
Table Page
2.1. Harvesting treatment. Areas of harvested and unharvested watersheds are shown in hectares (ha), total stream length within each watershed is given in meters (m), area of watershed harvested is given in hectares and percent of total watershed area, harvested stream length is given in meters and percent of total watershed stream length. .........24
2.2. Harvested-unharvested stream pairings for regression analysis. ........... 29
2.3. The warm season was divided into the following eight periods which were analyzed individually in the diel stream temperature analysis. ......................................................................................................... 31
2.4. A list of correlation coeffiecients between maximum, minimum and mean daily stream temperatures observed in harvested and unharvested streams...................................................................................... 35
2.5a: Differences between pre-harvest mean slopes and post-harvest slopes of daily maximum stream temperature regressions for each individual stream pair and overall. .................................................................. 37
2.5b: Differences between pre-harvest mean intercepts and post- harvest intercepts of daily maximum stream temperature regressions for each individual stream pair and overall. .................................................... 37
2.5c: Differences between pre-harvest mean slopes and post-harvest slopes of daily minimum stream temperature regressions for each individual stream pair and overall. .................................................................. 38
2.5d: Differences between pre-harvest mean intercepts and post-harvest intercepts of daily minimum stream temperature regressions for each individual stream pair and overall. .................................................... 38
2.5e: Differences between pre-harvest mean slopes and post-harvest slopes of mean daily stream temperature regressions for each individual stream pair and overall. .................................................................. 38
LIST OF TABLES (Continued)
Table Page
2.5f: Differences between pre-harvest mean intercepts and post- harvest intercepts of mean daily stream temperature regressions for each individual stream pair and overall. ......................................................... 39
2.6. Mean percent change in diel temperature fluctuation after harvesting in four harvested streams. Change is significant in every period except for June 1 to June 14. .............................................................. 46
2.7. Differences between mean pre-harvest annual maximum seven-day mean stream temperatures and post-harvest annual maximums in each stream. Myers and DeMersseman are unharvested............................. 48
2.8. Percent canopy closure and standard deviation in each surveyed reach before and after harvest. Fenton US, Clay US, Russell US and BB US were harvested in 2005. Clay DS was harvested in 2001. ................ 54
LIST OF APPENDIX TABLES
Table Page
A1. Regression line parameters for maximum daily stream temperatures in all stream pairs. .....................................................................91
A2. Regression line parameters for minimum daily stream temperatures in all stream pairs. .....................................................................92
A3. Regression line parameters for daily mean stream temperatures in all stream pairs. ...............................................................................................93
The influence of contemporary forest harvesting on summer stream temperatures in headwater
streams of Hinkle Creek, Oregon
Chapter I: Introduction
Justification
Commercial forestry is a principal industry in Oregon and throughout
the Pacific Northwest. Currently, Oregon has 28 million acres of land
designated as forestland and 85,600 Oregonians are employed in the forestry
industry (Oregon Forest Resources Institute 2006). The income generated
and jobs supplied by the forestry industry are crucial to the economy of the
state of Oregon. However, the forestlands of the Pacific Northwest support
multiple uses in addition to timber, including recreation, high quality water
resources, and habitat for terrestrial and aquatic wildlife. Intensive forestry
operations may degrade the suitability of these lands to provide some
beneficial uses. In an effort to minimize the environmental impact of
commercial forestry on the landscape, the State of Oregon enacted the
nations first Forest Practices Act in 1971 to regulate forestland management.
Since the Oregon Forest Practice Rules have been in effect, considerable
resources have been directed to exploring procedures that lessen the impact
of forest operations on Oregons waterways while maintaining economically
sustainable harvest practices.
In recent years, populations of native anadromous salmonids have
been listed as federally Threatened or Endangered according to the national
Endangered Species Act. Declines in populations of anadromous salmonids
are correlated with habitat degradation associated with intensive forest
management and stream temperature changes that occur in response to
management of surrounding watersheds may adversely impact aquatic habitat
for anadromous salmonids. However, the mechanisms and processes that
influence reach-level stream temperature patterns are not completely
understood and there is a need for data on the stream temperature effects of
2
contemporary forest harvesting on privately owned, intensively managed
forestland. The objectives of the following work are to
1. observe and quantify how stream temperatures in small, non-
fish-bearing headwater streams respond to contemporary
intensive harvesting practices, and
2. explain reach-level stream temperature responses through
investigation of pre- and post-harvest canopy closure.
Literature review
Physical controls to stream temperature
Observed stream temperatures are the result of interactions between
external sources of available energy and water and the in-stream mechanisms
that respond to and distribute the inputs of energy and water from external
sources (Poole and Berman 2001). Within Poole and Bermans categorization,
external stream temperature drivers are defined as processes or conditions
that control the relative amounts of energy and water that enter or leave a
stream reach. Available incoming solar radiation and water from upstream,
tributaries, or subsurface sources are examples of external stream
temperature drivers. Conversely, characteristics inherent to the streams
physical structure and the near-stream environment exert an internal control
on the stream temperature response to external inputs of heat and water.
Stream shading, channel morphology, and substrate condition are examples
of internal temperature controls.
The sources of heat energy exchange between a stream and the
surrounding physical environment can be summarized by the following model:
H N E C S A= in which H is the net heat energy gained or lost from the stream, N is heat
exchanged by net radiation, E is heat exchange from evaporation or
condensation, C is heat conducted between the stream water and substrate, S
is heat convected between the stream water and air, and A is advection of
3
incoming water from tributaries or subsurface sources (Moore et al. 2005,
Johnson and Jones 2000). The net radiation term in the energy balance
encompasses both inputs of shortwave (solar) and longwave (thermal)
radiation less emissions of longwave radiation. The input of shortwave
radiation is the only heat exchange process within the stream energy balance
that is unidirectional; shortwave radiation is delivered to the stream in the form
of solar energy but there is no mechanism for emission of shortwave radiation
(Boyd and Kaspar 2003).
The primary external driver controlling stream temperature is the
amount of solar radiation to which a stream is exposed (Brown 1969, Beschta
et al. 1987, Johnson and Jones 2000, Johnson 2004). Browns 1969 study
demonstrated that temperature change in stream reaches that receive little to
no advective input from groundwater sources can be predicted using an above
ground energy balance approach. Within the energy balance, the incoming
solar radiation term dominates the convective and evaporative components of
the model, and thus has the greatest impact on the amount of energy available
to the stream. Streams that are shaded, such as those that flow through intact
forests and are covered by the canopy, receive less solar radiation than
streams that are unshaded However radiation has the largest magnitude of
any term in the energy balance model, even in a fully shaded stream (Figure
1.1).
The relative effect of available solar energy on stream temperature
depends on the extent that solar radiation reaches the water surface. Material
that shades the stream controls the amount of solar energy that reaches the
stream surface by attenuating and reflecting solar radiation. Shade may be
provided by over- or understory riparian vegetation in any stage of life or
senescence. Topographic features or stream morphology and orientation may
also affect a streams exposure to solar radiation.
4
a b
Figure 1.1 Daily patterns of net radiation (Nr), evaporation (E) and convection (H) for a shaded (a) and unshaded (b) stream (Brown 1969).
The absolute amount of solar radiation that reaches a stream is only
part of the mechanism by which stream temperatures are raised. The surface
area and discharge of a stream are two additional factors that determine the
extent to which the temperature of a stream will fluctuate in response to
available solar radiation (Brown 1983). As the volume of water to be heated
increases, the effect of a fixed amount of solar radiation becomes diluted and
a smaller change in temperature is observed. Therefore, as stream discharge
increases, the increase in stream temperature associated with a given amount
of solar energy decreases. Conversely, as stream surface area increases, the
amount of solar radiation that the stream can absorb also increases, which
results in high net absorption per unit volume by a stream with a high surface
area to volume ratio.
Some researchers have stated that convective heat exchange is a
dominant process by which streams heat or cool (Larson and Larson 2001,
Smith and Lavis 1975). However, because air temperature and solar radiation
5
are highly correlated, it is often mistakenly concluded that air temperature
controls stream heating when, in fact, it is radiative exchange driven by
incoming solar radiation that causes stream temperature to increase (Johnson
2003). Energy balance analyses show that the magnitude of the incoming
solar radiation term is considerably greater than the convective heat exchange
term in the stream heat balance (Figure 1.1), (Brown 1969, Johnson and
Jones 2000, Sinokrot and Stefan 1993).
Substrate type affects the way a stream absorbs solar energy. Johnson
[2004] observed significant differences in maximum and minimum daily stream
temperatures as well as daily stream temperature fluctuations when a bedrock
reach was compared to an adjacent alluvial reach. Bedrock substrates of
small, shallow streams can absorb radiant solar energy, thus becoming energy
sources or sinks depending upon time of day. This process of absorption and
storage can dampen the diel temperature signal by storing or releasing energy,
resulting in lower maximum and higher minimum temperatures (Brown 1969).
However, Johnson [2004] found that a bedrock reach had wider diel
fluctuations than an alluvial reach, which suggests that the amount of solar
energy absorbed by the bedrock during the day and released at night was not
sufficient to dampen the diel fluctuation, as predicted by Brown [1969].
Furthermore, a dampening effect was observed after the stream flowed
through the alluvial reach. The increased residence time of water within the
alluvial reach may have allowed for conduction of heat between the surface
water and the alluvial substrates, thereby cooling warmer water during the day
and warming the cooler surface water at night.
Variable hydraulic residence times of individual streams may be
instrumental in producing divergent temperature responses across streams
that exhibit similar surface area to volume ratios and shade levels, and that
are exposed to comparable levels of solar radiation. The degree that surface
stream water interacts with the subsurface hyporheic zone can dramatically
influence hydraulic residence times (Boulton et al. 1998, Morrice et al. 1997,
Haggerty et al. 2002) and thus, temperature patterns within the surface water
6
column (White et al. 1987). Streams characterized by high surface-hyporheic
connection and long subsurface flowpaths may effectively thermoregulate
through natural heat-exchange processes as warm surface water mixes with
cooler subsurface water and remains in contact with subsurface alluvium
(White et al.1987). Morrice et al. [1997] illustrated that hydraulic residence
time increases with increasing hydraulic connection between surface
flowpaths and the subsurface alluvial aquifer. Using both point-specific tracer
analysis and reach-scale modeling, Morrice et al. [1997] demonstrated that
surface-hyporheic interaction is controlled by hydrogeologic attributes of the
channel substrate and the alluvial aquifer. Hydraulic conductivity of the
substratum, the magnitude and orientation of hydraulic gradients, stream
gradient and geomorphology and stream stage are physical variables that
influence rates and volumes of surface-hyporheic exchange (Morrice et al.
1997, Haggerty et al. 2002). In streams examined by Morrice et al. [1997],
substrates characterized by high hydraulic conductivities facilitated surface-
hyporheic exchange, resulting in greater hydraulic residence times through a
reach.
Though many studies and models agree that stream reach
temperatures increase in response to land use activities that enhance a
streams exposure to solar radiation, there have been disparate conclusions to
questions of downstream heat propagation and associated cumulative
watershed impacts. With regard to an above-ground energy budget, the
relatively diminutive magnitude of terms that could dispel heat (convection,
conduction and evaporation) as compared to the incoming solar term is
substantial. Solar radiation absorbed by a stream will result in an increase in
stream temperature but the increase will not be easily dissipated by
convection, conduction, and evaporation and therefore, theoretically, the
stream will cool more slowly than it is heated (Brown 1983). There is
ambiguity within current literature regarding what happens to stream
temperature downstream of a reach that was warmed by inputs of solar
radiation. Beschta and Taylors [1988] thirty-year study of stream temperature
7
and logging activity in the Salmon Creek watershed documents a significant
relationship between stream temperature at the mouth of the watershed and
cumulative harvesting effects which indicates that reach-level stream
temperature increases are detectable downstream. Oregon Department of
Forestry monitoring reports of the Brush Creek watershed indicate that stream
temperatures heated as the stream flowed through a clearcut reach but then
cooled so that there was no net heating observed at the watershed mouth
(Robison et al. 1995, Dent 1997). A Washington study that focused on
downstream effects of elevated temperatures in small streams concluded that
temperature increases in small streams were mitigated within 150 meters of a
confluence with a larger stream, however results varied from site to site
(Caldwell et al. 1991). Finally, Johnson [2004] demonstrated that maximum
temperatures in an exposed stream reach were cooler after the stream flowed
through a 200-meter shaded section than before the stream entered the
shaded section. The results of these studies signify that in some situations
stream temperature downstream of a disturbance is able to recover somewhat
more rapidly than is predicted by an above-ground energy balance but that the
temperature response downstream of a heated reach is variable.
The primary process of energy dissipation within a stream is generally
through evaporative heat flux, followed by emission of longwave radiation
(Boyd and Kaspar 2003). While rates of longwave radiation emission are
influenced only by water temperature, evaporative flux is controlled by
conditions in the near-stream environment. Vapor pressure gradients at the
air-water interface drive evaporation rates and so climatic conditions such as
humidity and windspeed significantly affect rates of evaporative flux (Benner
1999, Boyd and Kaspar 2003, Dingman 2002). Gauger and Skaugset
observed rates of evaporative heat flux on the order of 400 W/m2 in a stream
in the western Cascades of Oregon, and observed that wind enhanced rates
of evaporative heat flux (Gauger and Skaugset, unpublished data). While
most heat dissipation through evaporative heat flux occurs during the day
when humidity gradients between the stream and air and wind speeds are
8
greatest, net longwave emission away from the stream occurs at night when
stream temperatures become warmer than air and sky temperatures.
Physical effects of stream temperature
Maximum annual stream temperatures lag nominally one to two months
behind the time of annual maximum solar insolation (Beschta et al. 1987),
however, the timing of maximum annual temperature may change when
riparian vegetation is removed. Johnson and Jones [2000] report that streams
with disturbed riparian canopies reached summer peak temperatures close to
the time of maximum solar insolation despite the fact that stream discharge
was still high at that time while nearby streams with undisturbed riparian
canopies reached peak temperatures later in the summer. This observation
reinforces the dominance of solar radiation in determining stream temperature.
Aquatic organisms utilize dissolved oxygen (DO) for respiration for at
least a portion of their life cycle; thus DO concentration is a water quality
parameter of high significance to aquatic ecosystem health and is regulated
under the federal Clean Water Act. The solubility of oxygen decreases in
water as temperature increases; thus DO concentrations decrease as water
temperature increases. This relationship creates a direct link between water
temperature and quality of aquatic habitat. DO is consumed as organic matter
within the stream is oxidized by chemical and biological processes during
decomposition (Berry 1975, Ice and Brown 1978). Decomposition of organic
matter that is dissolved or suspended in the water column or associated with
the stream benthos contributes to a streams biological oxygen demand (BOD).
Rates of leaching, decomposition and associated BOD increase as water
temperature increases (Berry 1975). The addition of organic matter to
headwater streams in the form of logging slash contributes significantly to the
BOD of the system, dramatically reduces surface and intergravel DO
concentrations and may cause fish stress and mortality (Moring and Lantz
1975, Berry 1975).
9
Streams depleted of DO reaerate as oxygen from the atmosphere
diffuses into the water (Ice and Brown 1978). Reaeration through oxygen
diffusion occurs at the water surface and is enhanced by turbulence of the
water. Turbulence at the water-air interface entrains air into the water column
and brings oxygen-depleted water to the surface where it can reaerate (Ice
and Brown 1978). The rate of intergravel reaeration is low in comparison to
surface reaeration because the rate of water flux through benthic sediments is
much lower than stream velocities (Brown 1983, Berry 1975). Salmonids
begin their life cycle in redds as eggs and alevins that inhabit interstitial
spaces within streambed gravels and low intergravel DO levels can reduce
their survival (Ringler and Hall 1975).
Ecological effects of stream temperature
Water temperature criteria for streams in the Pacific Northwest were
developed to protect aquatic habitat for native, cold-water species, particularly
salmonids (Sullivan et al. 2000). Anadromous salmonids spawn and rear in
freshwater streams and resident salmonids fulfill their entire life cycles within
freshwater streams (Everest 1987). Therefore, the thermal environment of a
stream constitutes a vital metric of habitat quality that may determine the
ability of a stream to support salmonid populations. A shift in thermal patterns
of a stream may affect fish populations that are adapted to existing local
conditions, either through direct physiological pathways or by indirectly
modifying environmental conditions.
Stream temperatures that are sub-optimal can cause outright salmonid
mortality or may impose nonlethal effects that influence salmonid growth,
behavior (migration and reproduction) and pathogen resistance (Sullivan et al.
2000). The net effect of both lethal and nonlethal impacts to salmonid
populations depends on a combination of the severity and duration of
exposure to sub-optimal temperatures. Mortality occurs when either the
threshold magnitude or duration of extreme temperature exposure is exceeded.
Acute temperature effects include those that cause death after an exposure
10
time of less than 96 hours. Water temperatures over 25C generally exceed
maximum lethal temperature limits of salmonids (Brett 1952), although fish
that have acclimated to warm temperatures may persist above this threshold
for short periods of time (Brett 1956).
Chronic exposure to sublethal stream temperatures causes stress to
salmonids that is manifested through multiple physiological and behavioral
pathways and decreases the probability of salmonid survival (Elliot 1981,
Sullivan 2000). Physiological responses to a range of elevated but sublethal
temperatures indicate that while rates of some physiological functions such as
metabolic rate and heart rate increase continuously with increasing
temperature, other physiological functions such as growth rate and appetite
increase with temperature to a specific threshold, beyond which function
declines (Brett 1971). The development of a salmonid at the beginning of its
life cycle from egg to alevin, to fry and smolt occurs entirely within freshwater
streams and the rate of development at each life stage is largely controlled by
stream temperature. Stream temperature controls embryonic growth rates,
hatching time of embryos, time spent in the gravel of redds as alevin, and
emergence times and growth rates of fry (Marr 1966, Brett 1969, Weatherley
and Gill 1995). Growth rates of individual fry are determined by a balance of
energy expended by metabolism, activity and excretion to energy obtained
through food consumption. After basic survival demands are met, energy that
remains is applied to growth and reproduction (Brett 1969, Sullivan et al. 2000).
Brett [1969] related the variables of temperature and food consumption to
growth rates of salmonid fry and determined that the optimum growth rate for
all levels of food availability occurs at temperatures between 5-17C.
Maximum growth rates occurred at 15C when excessive food was available,
however temperatures for optimum growth decreased with decreasing food
availability and no growth occurred at temperatures above 23C. Growth rates
of fry influence survival and success in later life stages of development and
may determine the amount of time a fry of an anadromous salmonid will spend
11
in the stream before smolting and seaward migration occur (Quinn and
Peterson 1996, Weatherley and Gill 1995).
Water temperature directly influences salmonid behavior. Salmonids
may survive periods of exposure to sub-optimal temperatures by employing
behavioral thermoregulation and physiological energy-saving mechanisms
(Elliot 1981). Evidence of bioenergetic regulation of salmon fry in thermally
stratified lakes demonstrates that although many physiological processes are
maximized at 15C in the laboratory, under field conditions during times of low
food availability, salmonids naturally prefer cooler ambient temperatures
where maintenance metabolism is reduced (Brett 1971). Thermal
heterogeneity within a stream occurs when cooler subsurface water enters the
stream by subsurface seepage or hyporheic exchange, creating localized
areas of cooler habitat relative to the ambient stream temperature. There is
evidence that salmonids preferentially seek out thermal refugia during times of
temperature stress. Increasing frequency of pockets of cooler water is
positively correlated with increased salmonid abundance (Ebersole et al.
2003). Stream temperature also affects salmonid behavior during migrations
and thermal barriers to spawning adults may influence spawning locations and
migration timing (Lantz 1971).
An indirect effect of elevated stream temperature and increased
radiation is higher productivity of the stream ecosystem and a corresponding
increase in the availability of food, which has the potential to affect salmonid
populations. While the direct relationships between stream temperature and
salmonid health have been reasonably well observed and quantified through
laboratory experiments, defining comparable magnitudes of influence through
indirect pathways is a more challenging task due to the complexity of
ecosystem-wide relationships and challenges of performing ecological
research in-situ (Lee and Samuel 1976). In the Pacific Northwest, fish
communities are the highest trophic echelon of instream biota, thus fish are
indirectly influenced by changes in the productivity of lower trophic levels,
which include input of allochthonous organic matter, instream primary
12
production and aquatic invertebrates (Beschta et al. 1987). Water
temperature directly affects chemical and biological processes that occur
within the aquatic ecosystem, thus stream temperature is a ubiquitous control
to the productivity of the stream ecosystem. Stream temperature influences
rates of periphyton growth, organic matter decay and nutrient cycling by
controlling rates of chemical transformations within the water column, (Berry
1975, Phinney and McIntire 1965). Increases in stream temperature and light
availability that can result from forest harvesting may lead to shifts in biomass
production, species composition and dominance of algal communities within
the stream (Armitage 1980), which indirectly influences the trophic balance of
the stream. Studies that compared in-stream productivity in harvested and
unharvested streams often reported higher productivity in disturbed areas due
to increases in light and temperature (Murphy and Hall 1981).
Indirect linkages between water temperature and salmonid health exist
outside of the influence on food availability. The susceptibility of salmonids to
disease and parasites increases in warmer temperatures, presumably due to
the high metabolic rates and physiological stress associated with high
temperatures (Ordal and Pacha 1963, Cairns et al. 2005). Stream
temperature indirectly affects the quality of salmonid habitat by controlling the
solubility of oxygen in stream water. Salmonid mortality caused by low DO
concentrations occurs at concentrations less than 2mg/L, however nonlethal
impacts to salmonids are observed at DO concentrations as high as 6mg/L
(Hermann et al. 1962). Decreased growth rate, food consumption and food
conversion (weight gain) were observed in juvenile coho salmon when DO
concentrations decreased from 8.3 mg/L to 6 mg/L while mortality was
observed at 2.3mg/L (Hermann et al. 1962).
Aquatic insects fill a vital niche in lotic ecosystems by processing
organic material, thus providing a trophic link between primary production and
higher tropic levels. The preponderance of evidence in scientific literature
suggests that the instream thermal regime exerts a strong influence over the
aquatic insect community. Although laboratory studies that tested the lethal
13
limits of aquatic invertebrates showed that elevated or lowered water
temperatures induced mortality when lethal limits of a given species are
surpassed (Quinn et al. 1994), sublethal temperature effects may also
influence the life history patterns and overall long-term survivability of
macroinvertebrate populations. Water temperature affects the community
structure of aquatic invertebrates (Gledhill 1960, Hawkins and Hogue 1997)
and species extirpation was observed at temperatures above or below
threshold temperatures (Sweeney 1978, Quinn et al. 1994, Nordlie and Arthur
1981, Sweeney and Schnack 1977). Peak macroinvertebrate densities and
biomass occurred earlier in streams heated above ambient temperatures
(Arthur 1982, Hogg and Williams 1996, Rogers 1980) and emergence of adult
insects were observed earlier in streams heated as little as 2.5 to 3C above
ambient temperatures (Nordlie and Arthur 1981, Hogg and Williams 1996,
Rempel and Carter 1987). Stream temperature also influences rates of growth
and affects reproductive success of aquatic insects. Temperature directly
controls the metabolic rate of a given organism (Gillooly et al. 2001), and thus
regulates the developmental rate of that organism (Rempel and Carter 1987)
and directly affects mature body size (Hogg and Williams 1996, Sweeney and
Vannote 1978, Sweeney and Schnack 1977). A compelling hypothesis that
relates macroinvertebrate growth to the thermal environment states that each
species has an optimal temperature regime that allows each individual to
reach a maximum adult size and fecundity and that subjecting a species to a
regime that is suboptimal (either warmer or cooler than optimal), results in
reduced adult size and fecundity (Sweeney and Vannote 1978, Vannote and
Sweeney 1980). This hypothesis is supported by data that demonstrate
reduced adult body size for aquatic insects raised at temperatures above
(Hogg and Williams 1996, Rempel and Carter 1987) and below (Sweeney and
Schnack 1977, Sweeney and Vannote 1978, Sweeney 1978) the ambient
thermal regimes as compared to populations raised within ambient
temperatures and by studies correlating adult body size to fecundity (Rogers
1983, Sweeney and Vannote 1978, Hogg and Williams 1996).
14
Stream temperature and forestland management
The relationships between streamflow, solar radiation, shade and
stream temperature are prominent in the Pacific Northwest, where intensively
managed forest land and streams that support an economically, culturally and
ecologically valuable salmon fishery coexist. Incoming solar radiation peaks
during the summer months of May, June, July and August. Paradoxically,
climate patterns in the Pacific Northwest result in low probabilities of rainfall
and high probabilities of clear skies during the summer months, with the result
that peak annual solar energy is available during the times of lowest annual
stream discharge (Beschta et al. 1987). Small, headwater streams in the
Pacific Northwest are vulnerable to increases in temperature during summer
low flow months when incident solar radiation is high, particularly when
riparian vegetation is removed from streams that were historically shaded by
intact forest canopies.
Change to the thermal regimes of forest streams can be an undesirable
effect of vegetation removal within the watershed. The historic Alsea
Watershed Study demonstrated that the removal of streamside vegetation
during forest harvesting caused increases in stream temperatures (Brown and
Krygier 1970). Average monthly maximum stream temperatures increased
8C the summer after the forest adjacent to a small stream in Oregons Coast
Range was clearcut. In the same stream, diel stream temperature range
doubled after clearcutting. The importance of shade was further demonstrated
in Levno and Rothachers [1967] work in the HJ Andrews Experimental Forest
in western Oregon. Maximum weekly stream temperatures in a 96-hectare
watershed that was clearcut harvested did not diverge significantly from pre-
logging temperature patterns until 55% of the vegetation was removed from
the watershed. In the same study, no significant changes to stream
temperature patterns were observed one year after 25% of 101-hectare
watershed was patch cut. Downed wood and understory vegetation remained
near the stream in the patch-cut watershed the first year following harvesting,
however this material was removed during a winter debris flow that scoured
15
the channel to bedrock, exposing 1,300 feet of the channel to direct solar
radiation. Stream temperatures were significantly higher following the debris
flow than either before logging or one year after logging, which indicates that
the downed vegetation provided shade to the stream and precluded stream
temperature increases one year after logging. Brown and Krygier [1967]
quantified a 9C increase in stream temperatures as water flowed through the
1,300-foot reach that had been was scoured.
The role of senescing organic material as a temporary agent of shade
was defined in a study of headwater streams in western Washington (Jackson
et al. 2001). Post-harvest stream temperatures in headwater streams were
not significantly different than pre-harvest temperatures one year after the
streams were clearcut without a vegetated buffer. Jackson et al. [2001]
attributed the insignificant temperature response to the meter-thick layer of
organic material (logging slash) that covered the clearcut streams and
effectively excluded solar radiation after harvesting.
Increases to stream temperatures caused by forest harvest adjacent to
streams can be mitigated by Best Management Practices (BMPs), such as
retention of riparian vegetation on either side of a stream (Bescheta et al. 1987,
Brown and Krygier 1970, Brazier and Brown 1973, Macdonald et al. 2003,
Swift and Messer 1971). Gomi et al. [2006] reported increases in maximum
daily stream temperature of 2-9C in unbuffered headwater streams while
maximum daily temperatures in streams with 10- and 30-meter buffers did not
increase significantly. Similarly, the temperature increases observed in the HJ
Andrews and Alsea paired watershed studies occurred in streams where
riparian vegetation was clearcut or removed by debris flows whereas the
streams with intact riparian buffers did not warm significantly (Levno and
Rothacher 1967, Brown and Krygier 1970).
The characteristics that optimize effectiveness of riparian buffers have
been thoroughly studied are known. Brazier and Brown [1973] reported that
the volume of commercial timber left in the riparian buffer did not correlate with
the amount of energy deflected by the buffer but that the width of the buffer
16
(up to 40 feet) and canopy density of the buffer was directly proportional to
temperature protection. In an investigation of riparian temperature gradients
and edge effects, Brosofske et al. [1997] concluded that a minimum buffer
width of 45 meters was necessary to preserve an unaltered riparian
microclimate. In addition to length, width and basal density considerations, the
effectiveness of a buffer is directly related to its long-term stability. Macdonald
et al. [2003] reported that windthrow often occurs in riparian buffers and the
loss of canopy in years following harvesting inhibited stream temperature
recovery.
To minimize the environmental effects of forest harvesting on streams,
buffer rules were included in Oregons Forest Practices Act (OFP). Current
OFP regulations require forest operators to leave a buffer of riparian
vegetation or a Riparian Management Area (RMA) adjacent to streams that
support either populations of fish or a domestic use, or large and medium
sized streams that do not support fish or a domestic water use. The width of
the required RMA ranges from 6 to 30 meters from the stream, depending
upon beneficial use (domestic, fish, or neither) and size classification (small,
medium, large) of the stream. Within the RMA, forest operators are required
to retain:
1. a Standard Target square footage of basal area per 300 meters
of stream (basal area retention depends on stream use, stream
size, and silvicultural system),
2. all understory vegetation within three meters of the high water
level,
3. all overstory trees within six meters of the high water level,
4. all overstory trees that lean over the stream channel, and
5. a portion of live, mature conifer trees in the RMA (number of
trees retained depends upon stream use and size) (Oregon
Administrative Rule 629-635).
Rules regarding RMAs in other timber-harvesting states of the Pacific
Northwest are similar to the buffer rules mandated in Oregons Forest Practice
17
Rules. Like Oregon, California, Washington and Idaho designate varying RMA
widths and canopy densities depending upon stream size and beneficial use
(Adams 2007). Minimum RMA widths are greater for streams in Washington,
Idaho and California than for streams in Oregon. Additionally, Washington
designates a 15-meter core zone within the larger RMA for fish-bearing
streams in which no harvesting may occur. Portions of non-fish-bearing
streams in Washington, California, and Idaho that drain to fish-bearing
streams are protected by required RMAs of merchantable timber. In
Washington, the first 90-150 meters of perennial, non-fish-bearing stream
above a confluence with a fish-bearing stream is protected by a no-harvest
RMA while Idaho designates RMAs on the first 150-300 meters of non-fish-
bearing stream above a confluence. California mandates that RMAs of
overstory trees be retained on any stream that demonstrates aquatic life
(Adams 2007). In Oregon, RMAs of overstory conifers are not required
adjacent to small, non-fish-bearing streams that are not domestic water
sources. OFP Rules may require that all understory vegetation and non-
merchantable timber be retained within three meters of the stream depending
on the Geographic Region in Oregon that the stream is located and the size of
the watershed that the stream drains. In any case, small, non-fish-bearing
streams are not afforded the protection of a vegetated RMA that is designated
for larger streams.
There is concern that stream temperature increases that occur in these
unbuffered headwater tributaries may propagate downstream to larger, fish-
bearing reaches and that the combined impact of several warmed tributaries
may degrade aquatic habitat in fish-bearing streams. Since the OFP Rules
were first enacted, revisions have been made to update the Rules as the body
of knowledge regarding the impacts of forest management has expanded.
Recent recommendations by Oregons Forest Practices Advisory Committee
on Salmon and Watersheds (FPAC) include an extension of current buffer
rules to include a 15-meter RMA on either side of the first 150 meters of small,
non-fish-bearing streams above a confluence with a fish-bearing stream.
18
Within the 15-meter RMA, forest operators would be required to retain all non-
merchantable timber as well as four square feet of basal area per 30 meters of
stream. There is a need to determine what, if any, changes to stream
temperature are observed in small, non-fish-bearing streams in response to
current Forest Practice Rules and if impacts are observed, whether or not they
warrant a change in the current legislation.
19
Chapter II: The influence of contemporary forest harvesting on summer stream temperatures in headwater streams of Hinkle Creek, Oregon
Introduction
Stream temperature is a physical water quality parameter that directly
affects all aquatic life by controlling metabolism, growth, oxygen solubility,
organic matter decomposition and nutrient cycling within the stream
ecosystem (Phinney and McIntire 1965, Marr 1966, Brett 1969, Brett 1971,
Berry 1975, Weatherley and Gill 1995). Changes to prevailing thermal
regimes stimulate physiological and behavioral response mechanisms in
aquatic biota and effects ranging from physiological stress, changes in growth
rates, fecundity, trophic structure, competitive interactions and timing of life
history events and mortality are observed ecosystem responses to changes in
ambient water temperatures (Brett 1952, Brett 1971, Moring and Lantz 1975,
Sweeney and Vannote 1978, Beschta et al. 1987, Hogg and Williams 1996).
In extreme cases, changes to thermal characteristics may alter the stream
environment to the extent that native species are no longer able to inhabit their
historic range. Pacific salmonids are particularly vulnerable to increases in
stream temperature as they are cold-water fishes with lethal thermal tolerance
of approximately 25C that inhabit freshwater streams during almost every
stage of their life cycle (Brett 1952).
Many interacting mechanisms and processes contribute to observed
stream temperature patterns; however according to energy balance analyses,
solar radiation exposure is the primary temperature determinant of small,
shallow streams (Brown 1969, Johnson and Jones 2000, Johnson 2004).
Solar radiation exposure is limited by shade, such as from an intact forest
canopy, and extreme increases to reach-level stream temperatures have been
observed when forest canopies are removed (Levno and Rothacher 1967,
Brown and Krygier 1970, Swift and Messer 1971). Where Riparian
Management Areas (RMAs) that include mature timber are used, some
20
percentage of pre-harvest canopy closure is preserved and often significant
changes to stream temperature are not observed (Levno and Rothacher 1967,
Brown and Krygier 1970, Swift and Messer 1971, Macdonald et al. 2003, Gomi
et al. 2006). Recently the role of logging slash as an agent of post-harvest
shade has also been investigated. Jackson et al. [2001] attributed a damped
post-harvest temperature response of clearcut streams to exclusion of solar
radiation due to a thick layer of logging slash that was deposited over the
streams.
A key focus of contemporary watershed management is the role of
cumulative watershed effects from the summation of many seemingly benign
individual activities that produce a significant additive effect (Beschta and
Taylor 1988). Small, non-fish-bearing streams in some regions of Oregon do
not require that RMAs of overstory conifers be left during forest harvesting and
there is concern that reach-level stream temperature increases may propagate
into cumulative watershed effects, affecting downstream salmonid habitat. In
order to assess the likelihood of a cumulative watershed effect, it is important
to understand processes and mechanisms of stream thermal dynamics
operating at the reach scale. Considerable research has focused on the
effects of forest harvesting on stream temperatures, however, much of the
prominent research was done in the era of old growth conversion, using
equipment and techniques that were replaced by modern practices and before
the current suite of forest practice rules were put into place. An investigation
of the effects of timber harvest on stream temperatures on privately owned,
intensively managed forest land with young, harvest-regenerated forest stands
harvested using contemporary forest practices is necessary to assess reach-
level impacts of current practices.
The objectives of this study are to 1) identify and quantify changes that
occur to stream temperatures directly downstream of harvested units the first
summer after harvesting and 2) explain the stream temperature response by
examining differences in solar radiation exposure pre- versus post-harvest. I
hypothesize that the harvesting treatment will reduce canopy closure over the
21
harvested streams and that the increased exposure to solar radiation will
cause stream temperatures to become warmer after harvest.
Methods
Site description
This research was undertaken as part of the Hinkle Creek Paired
Watershed Study in association with the Watersheds Research Cooperative.
We examined the headwater streams of Hinkle Creek, a tributary to
Calapooya Creek that drains into the Umpqua River. The Hinkle Creek basin
is located in the western Cascades of southern Oregon, approximately 25
miles (40 kilometers) northeast of the city of Roseburg in Douglas County.
The Hinkle Creek watershed is comprised of two fourth-order stream
basins, the North Fork (basin area 873 hectares) and the South Fork (basin
area 1,060 hectares). The streams flow approximately southwest and
northwest, respectively, before they reach a confluence at the western
boundary of the study area. The elevation of the study area ranges from
about 400 meters above mean sea level (msl) at the mouth of the watershed
to about 1,250 meters above msl near the eastern boundary of the watershed.
Mean annual precipitation ranges from 1,400 mm at the mouth of the
watershed to 1,900 mm at the eastern divide.
22
Figure 2.1 Hinkle Creek study area. Black points represent approximate locations of temperature data loggers, flumes, transition points between fish-bearing and non-fish-bearing streams and downstream limits to timber harvesting.
The vegetation in the Hinkle Creek basin is dominated by harvest
regenerated stands of 55-year old Douglas fir (Pseudotsuga menziesii).
Riparian vegetation is comprised of understory species such as huckleberry
(Vaccinium parvifolium) and sword fern (Polystichum munitum), and overstory
species such as red alder (Alnus rubra). The fish-bearing reaches of Hinkle
23
Creek contain resident cutthroat trout (Oncorhynchus clarki). Roseburg Forest
Products (RFP) owns almost the entire watershed and the land is managed
primarily for timber production. Before the commencement of the Hinkle Creek
study in 2001, approximately 119 hectares of forest in the South Fork basin
(11% of the South Fork Basin) was harvested in three clearcut harvest units
(Figure 2.1).
Study design
The experimental design of the Hinkle Creek stream temperature study
is a Before After Control Intervention (BACI) paired watershed study intended
to identify and quantify stream temperature responses to forest harvesting in
headwater streams. Six headwater watersheds were selected for study within
the Hinkle Creek basin; four harvested (treatment) watersheds in the South
Fork basin and two unharvested (control) watersheds in the North Fork basin
(Figure 2.1). These headwater watersheds comprise the experimental units of
the presented research and will be the focus of the following work. The
orientation of the four treatment reaches in the South Fork basin is primarily
south-north while the two control reaches in the North Fork basin flow
approximately from west to east. Thirty-five hectares of the 2001 harvest units
fell within the South Fork headwater watersheds investigated in this study.
Four hectares (4%) of the Russell Creek watershed and 31 hectares (28%) of
the BB Creek watershed were included in the 2001 harvest units (Figure 2.1).
Each of the six headwater streams were instrumented with Montana flumes
and stream temperature data loggers at the approximate transition point
between a non-fish-bearing and fish-bearing stream designation so that
stream reaches upstream of the flumes are designated as small, non-fish-
bearing streams.
Harvesting treatment
Between July 2005 and March 2006, vegetation was harvested from the
four South Fork watersheds while the watersheds of the North Fork remained
24
unharvested. Harvest units were clearcut according to Oregons Forest
Practice Rules using modern harvesting techniques appropriate for each site.
Most harvest units were yarded using a skyline logging system, however a
portion of the harvest unit in the Fenton Creek watershed was shovel logged.
Felled trees were yarded tree length to the landing where they were processed
and removed from the project site via log trucks.
Table 2.1. Harvesting treatment. Areas of harvested and unharvested watersheds are shown in hectares (ha), total stream length within each watershed is given in meters (m), area of watershed harvested is given in hectares and percent of total watershed area, harvested stream length is given in meters and percent of total watershed stream length.
Watershed Name
Harvested/ Unharvested Watershed
Area (ha)
Stream Length
(m) Area Harvested
(ha, percent)
Harvested Stream Length
(m, percent)
Fenton Creek Harvested 20 900 15, 75% 620, 69%
Clay Creek Harvested 70 2,040 25, 36% 780, 38%
Russell Creek Harvested 100 1,800 10, 10% 630, 35%
BB Creek Harvested 110 2,280 35, 32% 1,060, 46%
Harvested Total 300 7,020 85, 28% 3,090, 44%
Myers Creek Unharvested 90 2,100 ----- -----
DeMersseman Creek Unharvested 160 1,580 ----- -----
Unharvested Total 250 3,680 ----- -----
The lower boundaries of the four harvest units coincided with the
locations of Montana flumes, the point where the streams transitioned
between a non-fish-bearing designation and a fish-bearing designation.
Therefore, all stream reaches located within the harvest units were classified
as small, non-fish-bearing reaches and according to the Oregon Forest
Practice Rules, a Riparian Management Area (RMA) of merchantable timber
was not required between the stream and harvest unit. Almost all
merchantable timber and most non-merchantable timber and understory
riparian vegetation was removed from riparian zones during harvesting.
Logging slash, consisting of branches, needles and understory vegetation was
25
left in place and harvested streams were partially covered by logging slash.
Site preparation for replanting began in Spring 2006 and included herbicide
treatments.
Stream temperature data collection
Summer stream temperatures in the six headwater watersheds were
monitored over a four-year period of calibration data collection (2002 through
2005) followed by one year of post-harvest data collection (2006). Average
stream temperature was recorded over 10 to 30 minute intervals using Vemco
12 bit Minlog data loggers (0.2C accuracy, used 2002 and 2003), or HOBO
Water Temp Pro data loggers (Onset HOBO model H20-001, 0.2C accuracy,
used 2004 through 2006). The data loggers were calibrated before
deployment to ensure accuracy between locations. HOBO or Vemco data
loggers were deployed each year in the late spring or early summer and
continuously logged stream temperature data until late fall. Data loggers were
located at the downstream edge of the proposed harvest units (Figure 2.1) and
were placed in the same specific locations each year. During post-harvest
data collection, data loggers were encased in white PVC covers to shade the
instruments from direct solar radiation. Holes were drilled in the PVC cases to
ensure that water flowed freely over the data loggers. Year-round stream
temperatures were recorded within 10 meters of each seasonal data logger at
30 minute intervals (Campbell Scientific CS547A conductivity sensors 0.1C
accuracy, used November 2003 through 2006).
Canopy closure data collection
Surveys of canopy closure over the gauged streams were taken during
the summer of 2004 and repeated during the summer of 2006. In this study,
canopy closure is defined as the proportion of sky that is covered by
vegetation that attenuates solar radiation before it reaches the stream
(Jennings et al. 1999). The four harvested streams were surveyed at ten-
meter intervals from a distance of 300 meters downstream of the downstream
26
limit of the proposed harvest boundaries (flumes) to at least the upstream
limits of the proposed harvest units (Figure 2.2). The unharvested streams
were surveyed at ten meter intervals from a distance of 300 meters
downstream from the flumes to at least 400 meters upstream of the flumes.
27
Figure 2.2. The locations of flumes and reaches surveyed for canopy closure in 2004 and 2006. The number of sampling points taken during the 2006 survey is displayed by each reach. The number of sampling points taken during the 2004 survey was equal or greater than the 2006 survey sample size for each reach.
Percent canopy closure was determined by measuring canopy closure
upstream, downstream, perpendicular to the stream on river right and
perpendicular to the stream on river left with a spherical densiometer held at
waist height. The four canopy closure measurements at each location were
averaged to calculate percent canopy closure at each sampling location. The
28
densiometer operator took canopy closure measurements from the center of
the stream.
During the summer of 2006, the percent canopy closure survey was
repeated to gather post-harvest data on levels of shading in harvested and
unharvested reaches. Percent canopy closure was sampled every ten meters
along each of the six streams using methods similar to the pre-harvest survey.
However, because the spherical densiometer held at waist height did not
adequately characterize shade provided by downed vegetation in the streams,
a second survey method was employed. Digital photos were taken at each
sampling location from a perspective of two to eight inches above the water
surface. Photos were taken close to the center of the stream at the exact
location of densiometer data collection. A bubble level attached to the camera
ensured that the photo captured a sampling area directly above the stream
and each photo was taken facing north. The photos were analyzed by
classifying proportions of light and dark pixels as canopy openness or closure,
respectively in Adobe PhotoShop 7.0 software.
Data analysis
Maximum, minimum and mean daily stream temperatures
Parameter analysis of regression curves was used to detect changes to
daily maximum, minimum and mean summer stream temperatures in Hinkle
Creek (Meredith and Stehman 1991, Loftis et al. 2001). All statistical analysis
was conducted within SAS version 9.1 (SAS Corporation, Cary, NC).
Maximum, minimum and mean daily stream temperatures were extracted from
the full temperature dataset of 10-30 minute observations and the three
temperature metrics were analyzed separately. In order to meet the
independence assumption inherent to regression, partial autocorrelation plots
were examined for data from each stream, each year to determine the time
period over which maximum daily temperatures were autocorrelated. This
analysis indicated that the maximum lag time between autocorrelated values
of daily maximum temperature was two days, thus a dataset consisting of the
29
daily maximum temperature of every third day was systematically selected
from the full dataset, with a randomly selected first day. Identical data
selection techniques were used to select an independent set of minimum and
mean daily temperatures. A two-day maximum lag time was identified for daily
minimum and mean stream temperatures and so the final independent dataset
also consisted of minimum and mean temperatures from every third day.
Examination of residuals reflected that all assumptions of regression were
adequately met by the data. Data from 2002 at Russell Creek were flawed
due to direct solar absorption by the data logger and so data from this stream
and year were removed from all analyses. Harvesting began in Fenton Creek
during the summer of 2005, thus all stream temperature data collected in 2005
in Fenton Creek were not considered in this analysis.
A set of geographic and hydrologic characteristics for each watershed
was considered to pair each harvested stream to an unharvested stream.
Average basin aspect, average stream orientation, stream length upstream of
the temperature sensors and stream discharge were considered in this
analysis, resulting in the following stream pairings:
Table 2.2. Harvested-unharvested stream pairings for regression analysis.
Harvested Stream Unharvested Stream Pair Name
Fenton Creek Myers Creek Fen
Clay Creek Myers Creek Clay
Russell Creek DeMerrseman Creek Rus
BB Creek DeMerrseman Creek BB
After watershed pairing was established, the daily maximum
temperatures from each harvested stream were plotted against daily maximum
temperatures collected on the same day from the paired, unharvested stream.
A Least Squares regression line was fit to data from each year, resulting in five
regression lines (four pre-harvest and one post-harvest) for each stream pair,
except for the Rus pair which lacked 2002 data from Russell Creek and the
Fen pair which lacked 2005 data from Fenton Creek. From each regression
30
line, a slope and intercept (C) parameter were extracted (Tables A1-A3).
Before regression lines were fit to the paired harvested-unharvested
relationships, the unharvested temperature data were adjusted by subtracting
the mean value of the annual means of daily maximum temperature (2002-
2006). This adjustment repositioned the scale of the x-axis, which allowed the
intercept of the regression line to fall in the mid-range of the observed stream
temperature values, precluding the need to extrapolate the intercept beyond
the range of observed data. Similar regression analyses were performed for
minimum and mean daily temperatures.
In order to detect changes between pre-harvest and post-harvest
slopes and intercepts of the regression relationships, the following repeated
measures model was fit to both the slope and intercept datasets:
= + S + Y I Y I Y I Y I
= slope / intercept for year i (i = 2002, 2003, 2004, 2005, 2006), stream pair j (j = Fen, Rus, Clay, BB)
overall mean slope / intercept for all stream pairs, all yearsS = random effect of stream pair that adds variability to the value of ,
j = Fen, Rus, Clay, BB; S ~ N(0, Y effect of year iI indicator; = 1 if 2002, 0 otherwiseI indicator; = 1 if 2003, 0 otherwiseI indicator;
ij 0 j i 2 i 3 i 4 i 5
ij
0
j
j S2
i
2
3
4
$
$
+ + + +
=
)
====
ij
= 1 if 2004, 0 otherwiseI indicator; = 1 if 2005, 0 otherwise
random error term that represents variability between years;
~ MN(0, ) and =
5
j
==
ij
5 5
2 3 4
2 3
2 2
3 2
4 3 2
11
11
1
An autoregressive (AR(1)) correlation structure between time periods is
the most appropriate correlation structure for repeated measures through time
and therefore was selected for this model. Examination of residuals confirmed
31
that the data adequately met all assumptions inherent to the model. Contrasts
between mean slopes and intercepts before and after harvest were used to
detect changes to the harvested-unharvested relationships of maximum,
minimum and mean daily temperature that occurred between pre-harvest
years and the post-harvest year.
Diel temperature fluctuation
Diel temperature fluctuation was calculated by subtracting the daily
minimum temperature recorded at each stream from the daily maximum
temperature. Diel ranges for every day between June 1 and September 30
were considered in this analysis. As diel range tends to fluctuate in a natural
seasonal pattern throughout the summer, the season was divided into discrete
periods and analyzed separately (Table 2.3).
Table 2.3. The warm season was divided into the following eight periods that were analyzed individually in the diel stream temperature analysis.
Period Dates
1 June 1 to June 14
2 June 15 to June 30
3 July 1 to July 14
4 July 15 to July 31
5 August 1 to August 14
6 August 15 to August 31
7 September 1 to September 14
8 September 15 to September 30
Changes to diel range were detected by examining the diel range
relationship between harvested and unharvested streams before and after
harvesting. The pairing of harvested to unharvested streams employed in the
maximum, minimum and mean analysis was also applied to diel analysis
(Table 2.2). Missing data were simulated by interpolating within regression
relationships between the HOBO temperature data logger at each site and the
Campbell Scientific temperature probe located on the adjacent flume. The
32
ratio of harvested to unharvested diel range was calculated for each stream
pair and a repeated measures model was fit to the diel range ratio dataset.
Examination of residuals indicated unequal variance, thus the natural log of
the harvested to unharvested ratio of diel range was used to correct for
heteroscadacity within the data. All other assumptions of the model were
adequately met by the data. The following repeated measures model was
used to detect changes to diel stream temperature fluctuation that occurred
after harvesting:
log( ) = + S + Y I Y I Y I Y I
) = logged ratio of harvested over unharvested diel range for year i (i = 2002, 2003, 2004, 2005, 2006), stream pair j (j = Fen, Rus, Clay, BB)
overall mean ratio for all stream pairs, all yearsS = random effect of stream pair that adds variability to the value of ,
j = Fen, Rus, Clay, BB; S ~ N(0, Y effect of year iI indicator; = 1 if 2002, 0 otherwiseI indicator; = 1 if 2003, 0 otherwiseI indicator;
ij 0 j i 2 i 3 i 4 i 5
ij
0
j
j S2
i
2
3
4
$
log( $
+ + + +
=
)
====
ij
= 1 if 2004, 0 otherwiseI indicator; = 1 if 2005, 0 otherwise
random error term that represents variability between years;
~ MN(0, ) and =
5
j
==
ij
5 5
2 3 4
2 3
2 2
3 2
4 3 2
11
11
1
An autoregressive (AR(1)) correlation structure between time periods is
the most appropriate correlation structure for repeated measures through time
and therefore was selected for this model. Contrasts between average diel
ratio before and after harvest were used to detect changes to diel temperature
range that occurred between pre-harvest years and the post-harvest year.
33
Greatest annual seven-day moving mean of the maximum daily temperature
Seven-day moving mean of the maximum daily stream temperature
(seven-day mean) was calculated for every day of the summer for each
stream, each year. The relationship of seven-day mean between harvested
and unharvested streams was used to assess changes to seven-day mean
that occurred after harvesting. The pairing of harvested to unharvested
streams used in prior analyses was used to assess changes to annual
maximum seven-day mean (Table 2.2). The maximum annual seven-day
mean of each unharvested stream was subtracted from the maximum annual
seven-day mean of the corresponding harvested streams. The following
repeated measures model was used to assess changes to the differences
between annual maximum seven-day means of harvested and unharvested
streams after harvesting occurred:
= + S + Y I Y I Y I Y I
= difference between harvested and unharvested 7 - day annual maximum for year i (i = 2002, 2003, 2004, 2005, 2006), stream pair j (j = Fen, Rus, Clay, BB)
overall mean difference for all stream pairs, all yearsS = random effect of stream pair that adds variability to the value of ,
j = Fen, Rus, Clay, BB; S ~ N(0, Y effect of year iI indicator; = 1 if 2002, 0 otherwiseI indicator; = 1 if 2003, 0 otherwiseI indicator;
ij 0 j i 2 i 3 i 4 i 5
ij
0
j
j S2
i
2
3
4
$
$
+ + + +
=
)
====
ij
= 1 if 2004, 0 otherwiseI indicator; = 1 if 2005, 0 otherwise
random error term that represents variability between years;
~ MN(0, ) and =
5
j
==
ij
5 5
2 3 4
2 3
2 2
3 2
4 3 2
11
11
1
34
An autoregressive (AR(1)) correlation structure between time periods is
the most appropriate correlation structure for repeated measures through time
and therefore was selected for this model. Examination of residua